Akademik Veri Yönetim Sistemi
63rd Conference on Decision and Control, Milan, İtalya, 16 - 19 Aralık 2024, ss.3378-3383, (Tam Metin Bildiri)
In this paper, we consider both the fixed-gain control and adaptive learning architectures to suppress the effects of uncertainties. We note that fixed-gain control provides more predictable closed-loop system behavior, but it comes at the cost of knowing uncertainty bounds. On the other hand, adaptive learning removes the requirement of this knowledge at the expense of less predictable closed-loop system behavior compared to fixed-gain control. To this end, this paper presents a novel symbiotic control framework that integrates the advantages of both fixed-gain control and adaptive learning architectures. In particular, the proposed framework utilizes both control architectures to suppress the negative effects of uncertainties with more predictable closed-loop system behavior and without the knowledge of uncertainty bounds. Both parametric and nonparametric uncertainties are considered, where we use neural networks to approximate the unknown uncertainty basis for the latter case. Several illustrative numerical examples are provided to demonstrate the efficacy of the proposed approach.